Field of the Invention
The present invention relates to an image capturing apparatus using an image sensor such as a CMOS image sensor, and a method of controlling the same.
Description of the Related Art
In recent years, an image sensor such as a CMOS image sensor is often used for a digital camera and video camera. Recent digital cameras as well as video cameras have a function of capturing and recording a moving image.
If an image sensor with more than 10 million pixels is used, when capturing a still image, it is possible to obtain a high-resolution image by reading out all the pixels of the image sensor. On the other hand, the number of pixels required for capturing a moving image is different from that for capturing a still image and, for example, the resolution of an HD format called Full HD is 1,920×1,080 pixels (about 2 million pixels). That is, to record a moving image using an image sensor with more than 10 million pixels, it is necessary to decrease the number of pixels.
Examples of a method of decreasing the number of pixels are thinning processing of skipping a pixel at a predetermined interval, addition processing of adding a predetermined number of pixel signals when outputting them from the image sensor, and crop processing of reading out only a specific pixel region.
These methods can be executed by an image processing unit or the like after reading out all the pixel signals from the image sensor, or executed within the image sensor or when reading out pixel signals from the image sensor. In the latter case, the number of pixels read out from the image sensor is smaller, and thus a data amount transferred from the image sensor is smaller. Therefore, the latter case is advantageous in increasing the speed of an operation of reading out the pixel signals.
As a method of performing addition processing for pixel signals within an image sensor, Japanese Patent Laid-Open No. 2010-034895 discloses a method of adding charges in a floating diffusion region (Floating Diffusion: FD) in the vertical direction, which is connected to photoelectric conversion units.
Furthermore, Japanese Patent Laid-Open No. 2007-173950 discloses a method of performing addition processing by simultaneously selecting a plurality of pixel rows by a row selecting circuit, and simultaneously outputting pixel signals on a plurality of rows.
As described in Japanese Patent Laid-Open No. 2010-034895, if the charges of pixels in the vertical direction are added in the FD, it is possible to obtain a high-quality image without moiré, but horizontal stripe noise is readily generated in an image due to noise via the parasitic capacitance of an addition switch. Especially when reading out pixel signals at a high gain to improve the sensitivity, not only the pixel signals but also noise is amplified. Therefore, as the sensitivity is higher, horizontal stripe noise is more conspicuous, thereby interfering with obtaining of high sensitivity.
As described in Japanese Patent Laid-Open No. 2007-173950, when mixing outputs of a plurality of pixels on a signal line, if the difference between the outputs of the pixels to be mixed is large, the mixed output is limited by an operation range, thereby degrading the image quality. In particular, imaging at a low gain is readily influenced since a large amount of charge is processed and thus the output difference between pixel signals is large.